Even 2020 cannot bring forth the Year of Linux on the Desktop
- Reference: 1602761663
- News link: https://www.theregister.co.uk/2020/10/15/even_2020_cannot_bring_forth/
- Source link:
Open source software advocate Eric S Raymond [2]prised the lid off the can of worms late last month, suggesting that Windows development was "a thing Microsoft would prefer not to be doing" while praising the likes of Windows Subsystem for Linux (WSL.)
Barnes has had a long association with WSL and was one of the movers and shakers behind the Pengwin Linux environment for Windows 10. [3]Pengwin itself was a rebrand of [4]WLinux which had the neat party trick of bringing Linux GUI apps to Windows 10 (X client required) before such shenanigans turned up in Microsoft Build demos.
Dismissing the idea of Microsoft shunting the core of the Windows OS to the Linux kernel as "a long-held fantasy for open source and Linux advocates", Barnes explained there were both product and simple engineering reasons why such a move was highly unlikely.
The main one is simple compatibility. Microsoft's own attempts to shift customers to a different platform have memorably failed in the past due to compatibility problems (yes, we're looking at you Windows RT.) Even it's flagship Surface Pro X 2-in-1 has struggled in part because its Windows-on-Arm approach does not yet support 64-bit Windows applications.
Getting to the level of compatibility to make such a rebasing possible "goes beyond contributing some fixes to Wine," explained Barnes. The experience of getting WSL to where it is today, where a jump from a translation layer to adding a virtualized Linux kernel to Windows was required in order to maximise compatibility. Attempting the level of compatibility needed for every Win32 API and syscall would be a huge undertaking.
"It is one thing to provide a NT-related compatibility layer for SQL Server on Linux, which Microsoft owns," stated Barnes, "it is another to guarantee it for millions of other developer's applications."
And as for the vast array of drivers enjoyed by the NT kernel, Barnes reckoned that shunting them into Linux would cost millions.
Furthermore, rebasing Windows on Linux could have the ironic effect of stifling competition should a kernel monoculture take hold. One need only glimpse at the Chromium-obsessed browser world to get a preview of the dangers that lie that way.
Barnes said he sees a future free of the desktop OS wars of old, where multiple operating systems co-exist: "The choice will not really be Windows or Linux," he said, "it will be whether you boot Hyper-V or KVM first."
Indeed, a more interesting question isn't so much the red herring of Windows shifting to Linux, but just how many secrets Microsoft is willing to open-source. Many Windows components (such as Terminal) already live in the open source world, and the direction of travel points to more following suit.
It is time to put the long-held dream of one OS squashing the other back into the cupboard of things we worried about in previous decades.
"The key take-away though," said Barnes, "is that open source has won." ®
Get our [5]Tech Resources
[1] https://boxofcables.dev/no-microsoft-is-not-rebasing-windows-to-linux/
[2] https://www.theregister.com/2020/09/28/eric_raymond_linux_beats_windows_prediction/
[3] https://www.theregister.com/2019/03/15/wlinux_becomes_pengwin/
[4] https://www.theregister.com/2018/09/26/wlinux/
[5] https://whitepapers.theregister.com/
Re: Is this really a valid point?
... if you don't run any Windows version later than 7 because you're sensible, it's hard already to find even a new printer with drivers to suit.
Surely, if you're sensible, the printer is just a device on the network. You send it a PostScript file and it prints it. What need is there for a driver?
(OK, there is a little bit more to it than that, but you only need one driver for all network-connected PostScript printers.)
PostScript: the only good thing that Adobe have ever done.
Re: Is this really a valid point?
"Buying new hardware for Windows systems is almost impossible unless you keep "upgrading" the OS with every new version, whereas most current Linux distros support plenty of "legacy" hardware. "
This is incorrect on every level.
Re: Is this really a valid point?
Firstly I'd like to point out what a super-informative and useful post you've written, with well-reasoned and well-elucidated arguments. Bravo.
Secondly, how's your Packard Bell Fastmedia remote control running on windows these days? Mine is still running great on Linux, 20+ years later. In fact I managed to score a spare one recently in case the one I've been using since ~1998 ever dies.
That's just one example of "legacy" hardware that hasn't been supported on windows since it switched to the NT codebase.
But one example is all I need to demonstrate that "incorrect on every level" is...uh...incorrect.
Denials are step one
Ending support for a small proportion of legacy software was never a problem. It had the bonus of forcing most people to buy the new version (and a few to switch to Linux). Lack of new drivers for old hardware has always been a solution for selling new hardware. So far the cost of modifying Windows has been less than the benefit of a Windows licence on all new laptops and desktops. Phones, laptops and consoles have been cutting into desktop sales. The increased useful life of laptops have pushed Microsoft into the software rental business.
Buying laptops and desktops with Linux preinstalled is getting easier. There must be a sufficient market big market already that it has more value than the cost of push-back from Microsoft. I expect that market to grow as fast as stalactites. Come back in five years and we will see how much money Microsoft is willing to spend on supporting the most troublesome legacy apps.
Monoculture: MacOS, BSD, a fork of Linux, Windows extended support for $100,000 per seat per year (per legacy app?).
I am shocked, shocked I say, that ESR was spouting bullshit.
I disagree with Barnes. MS has done it before when they had two OS lines in parallel, the old Win16/32 world of a GUI on top of DOS and the new world of NT. These parallel worlds (including different driver models) went on for quite some years (95, 98, ME vs NT3/4, 2K) until finally Windows XP (NT-based) was deemed "backwards compatible enough" to end this double effort. There were still fears about compatibility problems but it went pretty well in the end.
Maintaining a web browser costs a lot of money... so MS finally bowed and accepted Chromium as the base for their browser.
Maintaining a kernel also costs a lot of money so...
Given time, they can do the same thing over again. Not in 2022, probably not in 2025, but by 2030 Windows will be Linux-based.
> but by 2030 Windows will be Linux-based.
What would anyone have to gain from this event that is emphatically not going to happen, even?
Morer kissing and hugging between MS and Canonical...
...where will it all end?
Honestly, who cares about the Desktop? And I say this as someone who uses a desktop (or laptop) computer every day, but they’re increasingly anachronistic. Phones and tablets are the computers that people use the most, and Android is top dog in that space - which, last time I checked, has Linux underpinnings.
If we forget the nebulous argument about Linux on the desktop, and extend it to all Unix alikes, then this is even more true since iOS and MacOS are making healthy inroads to the Windows space as well - and ChromeOS looks like it’s picking up traction too.
But really. Who cares? Most people seem to do everything on the web these days, so if it has a browser then that’s all that really matters.
who cares about the Desktop?
People interested in using complex software who want an interface that isn't dogshit
Re: Phones and tablets are the computers that people use the most
Yes, and an excellent choice for developing professional software.
Office on Linux...
Now that would kill the OS for many more people than anything else.
Of course nowadays you can't buy office without paying for Windows anyway...
Oh for the days when you could buy the bits of the pie that you actually needed.
Good to hear that. As the 2nd para says, it would be a con of worms as I doubt Microsoft would have been able to resist throwing its weight about.
I would so love to go back to old skool on this. New Xbox running full Win10 vs new Ps running Linux. But Mum, Dad i can do my homework on it too. Let there be war.
"X client required"
I'm reasonably sure you meant "X server" here. The X clients are the applications themselves.
Afternoon nitpicking: ✔️
Is this really a valid point?
"...as for the vast array of drivers enjoyed by the NT kernel, Barnes reckoned that shunting them into Linux would cost millions"
But at least they wouldn't then continue to to be version-specific. Buying new hardware for Windows systems is almost impossible unless you keep "upgrading" the OS with every new version, whereas most current Linux distros support plenty of "legacy" hardware. For example, if you don't run any Windows version later than 7 because you're sensible, it's hard already to find even a new printer with drivers to suit.